IMEV or IPMEP number), word or phrase. One very useful and
innovative feature is searching the texts by the LALME dialects
identification (fig 9).
The Corpus of Middle English Prose and Verse can be browsed and
searched for words and phrases as a whole or within particular
works.
A proximity search for ‘experience’ within 80 characters of
‘auctoritee’ gives two matches both from the Canterbury Tales.
Whilst a simple search for the phrase ‘man of gret auctorite’ in the
entire corpus brings no results, a Boolean search for all three
words occurring together within a line or a paragraph brings
twenty nine results, all from prose, though none of the phrase
itself. Thirteen of these hits come from the Anthology of Chancery
English which shows that though ‘auctoritee’ itself was commonly
used, Chaucer’s phrase at the end of the House of Fame was not a
common legal (or other) set phrase. The disappointing absence of
‘man of gret auctorite’ is interestingly counterbalanced by the
presence of a ‘lady of gret auctorite’—a search which gives one hit
(fig 10).
Cetedoc’s Thesaurus
Formarum Totius
Latinitatis
Platform: Windows 3.1 or higher.
Requirements: CD-ROM drive essential.
Available from: Brepols Publishers NV, 68
Steenweg op Tielen, B-2300 Turnhout, Belgium.
Tel.: +32 (0)14 402 500. Fax: +32 (0)14 428 022.
http://www.brepols.com/
Price: BEF 75,000 single user. Network price on
request.
Cetedoc’s Thesaurus Formarum Totius Latinitatis (TF) describes itself as
a ‘Database for the Study of the Vocabulary of the Entire Latin
World’. It aspires to contain every attested form of every word in the
Latin language from Plautus to the present day, together with
information regarding the periods, authors, and works in which
each form is found, and the frequency with which it appears in
them. The package consists of a substantial printed version of the
database, a more detailed electronic version on CD, and a small
manual to accompany the retrieval software. The printed database
contains an introduction to the database as a whole, and both that
and the manual are written in French and English. The discussion
below will be of the electronic database, unless specified.
Content
Fig.10
Incomplete as the results of this small investigation may be, they
still prove useful. And the learning of all this and the searching itself
only occupied around twenty enjoyable minutes of my time. The
usefulness of searching the Corpus will increase proportionally to its
size and we hope that the collection will continue to grow.
Conclusions
The Middle English Compendium is major achievement and a
powerful resource of modern scholarship in Medieval English. Much
of its capacity comes from the searchability and interconnectivity of
its resources. The MEC is user-friendly, providing online help and
explanations, and is fast and easy to navigate. Search forms require
minimum typing, the results are clearly presented.
Access to the project is subscription-based whilst at the same time is
being continually developed. It functions successfully in its developing
state and some effort has been made to provide guidance to the
unfinished features. The resource remains open for potentially
endless development, and with so much achieved already, it promises
to continue being a tremendous contribution to Middle English
scholarship.
Elizabeth Solopova
esolopova@britishlibrary.net
Page 32 Spring 2000
Every text that has been used to prepare the database has been
categorised under six headings: title, author, Clavis Patrum Latinorum
number (where appropriate), size, century, and period. The latter
refers to one of the four ‘ages’ into which the TF divides the history
of Latin: antiquitas (from Plautus to the end of the second century
A.D.), aetas patrum (from Tertullian to the death of the Venerable
Bede in 735), medium aevum (from 736 to 1499) and recentior latinitas
(from 1500 to 1965, the last texts being those from the Second
Vatican Council).
Any one of these headings, or any combination of them, can be
used to query the database: thus as well as the number of
occurrences of the form ‘amor’ in all Latin, one could find the
frequency of the form ‘Roma’ in all first century BC authors other
than Virgil, or all forms that occur more than 3000 times in the
patristic period, or, should one have the inclination, all the forms
common to Cicero and St. Augustine but found nowhere else,
which are found only once in those authors, and begin with letter
‘t’ (there are four). Any result can then be displayed in progressively
greater detail (e.g. frequency in various periods, in various authors,
in various works), and in various orders (alphabetical, numerical,
normal and reverse). The most detailed information available
about any form is the title of the work or works in which it
appears: for its exact location within the work, or for the actual
text, one needs to look elsewhere (for example in the Thesaurus
Patrum Latinorum or the PHI CD).
A lot of thought has gone into the categorisation of the data: for
example, works of any author are distinguished as authentic,
dubious and spurious; the dating of a work to a century is categorised
as either secure, doubtful, ambiguous or a terminus ad quem; a title
Computers & Texts No.18/19
to the Latin-speaking world after the 2nd century AD. The first ten are
christi, idest, christum, apostolus, ecclesiae, augustinus, david, iesu,
iesus and ecclesiam, suggesting, we are told, a ‘world of explanation,
in which idest comes second’, and also a ‘world with the omnipresence
of the Old and the New Testament’.
Fig.1. The results of a search for ‘fulmen’, showing the
highest level of detail.
thought to be due to the author of a particular work is included as
part of that work, as opposed to those titles given to a work by others.
A lot of effort has also gone into the categorisation of the forms, and
perhaps one of the most powerful features of the electronic TF is its
ability to deal with the differing orthographies and orthographical
conventions (or lack of them) found in different authors and editors:
with the printed database, as with many programs that exist for
searching electronic databases, not only does one have to perform
separate searches for each variant spelling of a word, but one also
has to know what those variant spellings are. While this kind of
problem occurs to an extent in classical Latin (especially with
compounds such as adfero or affero), it is much more frequent in later
Latin, with letters being doubled, added, removed or replaced with
abandon (e.g. penitentia against paenitentia, nichil against nihil,
screpitu against strepitu etc.). The introduction to the printed database
gives examples of some of the most common ways in which spellings
can vary, and the list is quite substantial. In the electronic database,
however, all the hard work has already been done, and the variant
spellings of any particular form are all linked together. This means,
for example, that on consulting the TF one can discover that
abundantia is found also as abandantia, abundancia, habundantia,
and habundancia (this facility would make a handy tool in its own
right), and then search for all five forms (or any selection from the
five) simultaneously: clearly an extremely useful feature.
What is it useful for?
The uses to which this database could be put may be immediately
obvious to some, but others may perhaps ask ‘what (or who) is the
TF actually for?’. According to the notes accompanying the database,
the TF provides an ‘important source of information … for the
history of the Romance languages and of all the European languages
influenced by Latin’, and ‘an initial opening to all Latinity and thus
is a tool that brings hope’. Exactly what kind of hope it brings is open
to question, but the TF is certainly a useful tool for the study of
language, enabling the user to discover the words which were
common at any particular point in history, the words which were
rare, and how this changed over time. A good example of one of the
many kinds of analysis possible can be found in the introduction to
the printed database, where the reader is presented with a list of all
the forms that appear more than 1000 times, but do not appear in
Antiquitas: that is, a list of all the most common forms that are new
Computers & Texts No.18/19
It is hard to comment on the absolute usefulness of the TF, since it
covers such a large time-span, and is thus potentially of interest to
a number of different disciplines. However, it is clear that the TF is
quite a flexible tool, and it may have something to offer to everyone,
though of course exactly what it can offer and the resulting value for
money may differ considerably. For those such as myself whose
interest lies in Antiquitas, the TF has immediate advantages over the
PHI CD when it comes to investigating how common a particular
word is, or obtaining a list of forms unique to Vergil, for example, or
a list of forms common to Propertius and Tibullus (though this is not
the same as a list of words). The ability to search for variant spellings,
and the powerful search facilities of the TF in general (it supports
multiple individual and multi-character wildcards), are also
extremely useful, and can point to forms missed with a careless
search of the PHI CD. The database also has pedagogical potential:
one can soon discover the most common forms in any particular
author, or selection of authors, and use them as the basis for a
specialised vocabulary list. One can also satisfy one’s curiosity for
pleasing trivia such as the ten most common words in classical
Latin, which are (should anyone be interested) et, in, -que, est, non,
ut, cum, ad, quod, and si.
Help Documentation
My acquaintance with the TF began with the manual, which did not
prove to be an auspicious start. The manual is in two parts, the first
containing the original French version, the second an English
translation. Unfortunately, the translation is very poor, at times
unreadable. This is an example from the first page: ‘In the printed TF,
the researcher has the list of the real forms attested in the works that
means, the graphics such as they are printed in the editions’. Throughout
the manual, even simple features can be rendered unfathomable by
such oracular pronouncements as ‘With the “Jump” option you can set,
during the display, the movement in function of the serial number of the
displayed forms’, and although there are a large number of screenshots to illustrate the text, they do not make up for the deficiencies
of the translation. In general, the organisation of the material is
poor, and important information is frequently hidden in the midst
of seemingly unimportant text. A further problem is that the
manual nowhere explains or even acknowledges the existence of
two rather vital buttons (labelled Words and Lines) that are used for
almost every search: they allow the user to choose an author, title,
century etc. from a list, rather than type in the required form (e.g.
‘saeculum -ii [dubium]’) from memory.
The buttons are mentioned, however, in the on-line help, which is
available in English, French, German and Italian (one chooses a
language each time the program starts), and tends to be more
detailed than the manual. Help is available for each separate
screen, usually in the guise of a screen-shot, clicking on the buttons
of which reveals their function. There is a lot of useful information
in the on-line help, though again its organisation could be improved
with more cross-references and a better contents screen. My main
criticism, however, is that a lot of the important information
contained in the on-line help should really be in the manual (such
as the two missing buttons, or details of how to retrieve results that
Spring 2000 Page 33
one has saved), where it could be presented in a clear and logical
order, the screen-specific (and un-ordered) online help being better
suited to refreshing the memory. The manual does not provide a
good introduction to the software, especially for the many potential
users who are not at ease with technology, and who could easily be
discouraged from getting the most out of the powerful software.
While about an hour’s experimentation with the program, and
occasional consultation of the on-line help, can provide the user
with a good idea of how the TF works, such time is something of a
luxury, especially if the CD is going to be used in a library, and I do
not think I am alone in knowing a large number of academics who
do not like to experiment with computers.
Searching the Thesaurus
Again, first impressions were not favourable: although the
installation wizard successfully installed the program files to my
designated folder ‘C:\Classics Programs\CILF’, it was unable to
create a short-cut, the space between ‘Classics’ and ‘Programs’ in the
folder name causing trouble. While this kind of problem was
frequent in the early days of Windows 95, such a basic error should
not be found in software written in 1998.
Fig.2. Searching for Cicero.
Despite a number of other bugs and design infelicities, to which I
shall return later, the software does have its strong points. Once one
gets used to the slightly non-standard interface, which does not take
too long, it turns out to be very simple but very powerful, allowing
the most frequently performed tasks to be accomplished quickly and
easily. For example, supposing we wish to find the most common
words in Cicero: clicking on the Auctor button on the main screen
presents us with a new window (see Fig.2). The name of the author
can then be typed into the ‘search’ box, or selected from a list of
authors (the words button, unmentioned in the manual). Clicking
on the Next button reveals a list of Ciceros to choose from (authentic,
dubious, etc.). Once the selection has been made, clicking on the
Next button again allows the user to apply Boolean search terms to
the various selected authors, clicking on Next again allows the user
to specify a frequency: in this case we will be after forms that occur
more than three hundred times.
We are then returned to the main screen: here, should we wish,
the search can be modified still further: the TF allows these results
to be restricted to particular forms (e.g. forms that appear more
Page 34 Spring 2000
than 300 times, and end in -orum), or linked via Boolean terms to
any other search: so (for example) we could exclude from our
Cicero search the most common forms from Antiquitas, which
would exclude a lot of the ‘uninteresting’ forms such as et, non,
cum etc. from our results. The results are then displayed, and can
be saved for later use, or exported to a text file for use in another
program, or as a search argument for the Cetedoc Library of
Christian Latin Texts.
There are problems with the software, however: although the
results can be sorted by frequency, for some mysterious reason
they can only be sorted by their frequency in the entire database.
In order to sort by their frequency in Cicero, one has to export the
results to a text file, and then load them into a program such as
Excel or Word and do the sorting there. The program can also take
a long time to display the results, and during this time the
computer appears to freeze: even on a PII 333, a search yielding
a large number of results can lock up the computer for more than
a minute, leaving the user unable to switch to any other application
or do anything at all. Rather annoyingly, the TF does not support
long filenames, so that when saving results the user is forced to
return to the misery of obscure and unintelligible eight-letter
cryptograms. Having saved a number of results, it was then quite
difficult to retrieve them, as rather surprisingly one cannot load
them in from the File menu (one has to load them through the
Connections window): while the solution to this problem could be
found in the on-line help (nothing in the manual), it was not easy
to track down. As far as I am aware, the data cannot be copied to
the hard-drive to speed up performance, as the program always
reads from the CD.
A rather more serious problem occurred with the printing: although
the results were fine on one of the machines (running Windows
NT 4 and printing to an HP LaserJet 8000), on the other, running
Windows 95 and printing to a Canon Bubble-Jet 200ex, the text
appeared in tiny print (about 4 point size) and there were no
options to change it, thus making it impossible to obtain a hard
copy of one’s results without exporting them to a text-file and
then loading them in another program.
Conclusion
The TF presents itself very much as work-in-progress: the introduction
to the printed database acknowledges that given the ambitious
scope of the project there are bound to be errors in the information
contained within, and it stresses that this is the first of many releases.
Indeed, even the program-icon reads ‘TF1’, suggesting that there will
be a TF2 and a TF3 etc.. While I came across a number of minor errors
in the database (one example of many: facta appears 30 times in
Ovid’s Fasti, not 29), the frequencies were usually out only by one or
two, which would not affect broad surveys of the Latin language: it
was the manual and the software that seemed most in need of
attention, especially given the high price of the database. If one can
interpret the manual and ignore the bugs, however, the software is
powerful and versatile, and the thought and effort that has gone
into the preparation of the TF gives it the potential to be an
extremely useful research tool.
Matthew Robinson
Balliol College, Oxford
matthew.robinson@balliol.ox.ac.uk
Computers & Texts No.18/19